In pilot research using fMRI, we found that voluntary activation of primary motor cortical (Ml) foot/leg areas was retained in paraplegics (Ps) up to 10 yrs after injury, with overlap developing between Ml hand and foot areas (since confirmed by others). More extensive analyses of our scans from 14 Ps and 4 control subjects (Ss) have revealed, however, more striking changes in brain activation patterns than previously reported. Moreover, these changes develop progressively over 15-25 yrs after spinal injury. In control Ss, brain activation is well lateralized and confined to distinct areas during unilateral foot or hand movement. By 6 months to 2 yrs after spinal injury, however, attempts to move one foot now activates supplementary motor and premotor cortex, basal ganglia, and cerebellum bilaterally. By 3-6 yrs after injury, there is some return toward normal patterns of activation in the Ml and SMA, but greater activation now occurs bilaterally in the basal ganglia and in frontal premotor and association cortex. By 12-25 yrs after injury, these changes are much more pronounced. Ml regions previously active during attempted foot movement are now silent, as are previously activated regions of the posterior cerebellum. But bilateral activation increases further in basal ganglia and in premotor/association cortex, and spreads even to hippocampal, limbic, and parieto-occipital-temporal, "higher-order association areas". Thus, we find more global and progressive changes in brain activity when attempts are made to move a paralyzed limb than has been reported in previous studies, where all Ps were analyzed as a single group. But our sample is small. A new study is needed, therefore, in which larger groups with differing injury durations are studied for greater statistical reliability, and in which improved motor tasks and kinematic measures allow for better control of movement parameters and effort. This study will expand our knowledge of how the brain changes progressively over decades to altered communication with the sensorimotor periphery, and could aid in the intelligent design of future therapies.